Re: UKNM: Online surveys

[Silas Denyer]

----- Original Message -----

From: John McCarthy <

j [dot] mccarthyucl [dot] ac [dot] uk

>

To: <

uk-netmarketingchinwag [dot] com

>

Sent: 15 September 2000 11:41

Subject: Re: UKNM: Online surveys



>

> >We have recently placed an online survey, in a secure part of our

website,

> >which means limited access so only our clients can contribute.  What I'm

> >interested in is that as samples go '30 being minimum true, 100 being

okay

> >and then 500 being optimum' - is there any info of this kind with

'online'

> >samples?  Same or different?   And any stats on optimum % of respondents

> >from target sample?

>

> There is no straight forward answer to the question apart from - the more

> the better.

>

> t really depends on what you are using the survey for. If you are trying

to

> identify differences between different market segments then a minimum of

30

> in each group is a good rule of thumb. However, if the difference you are

> looking for is very small - then you will need a lot of people to identify

> the difference.

>

> The number of people is like the power of a lens. The more you have the

> more detail you will be able to see.

>


Very nice analogy!


> The main issue to be aware of with any kind of sampling (online and

> offline) is the population (or group of people) you are interested in.

Most

> online surveys are self selecting - ie. only certain types of people fill

> them in and they have certain opinions. The danger is that you will not

get

> a representative sample of all the visitors to your site.

>


I can only give you some examples from surveys we've recently run. For one

client, with a registered user population of c.2500 we obtained c.200

responses to an online user survey, but (perhaps different to your

approach?) prompted by a mailout to all users. The results certainly point

to us having polled only the most active users of the site, i.e. those

actually interested in what is being offered, and therefore one has to look

slightly "behind" the raw data to understand what can be done to increase

the level of activity of the remainder.


To validate the results, we also conducted c.200 telephone and online

anonymous surveys of the target audience group (estimated size c.200,000),

which meant that we could get a handle on whether or not the "active" users

were representative of the target audience. The power of this "lens" is

pretty good indeed: the BBC's bespoke poll a couple of nights ago on the

fuel crisis was estimated to have an accuracy of +/- 4% on a sample size of

c.500 people out of the entire adult population of the UK.


Although you may not be able to run to that sort of dedicated survey

activity, it is certainly worth getting demographic information from your

users, and then trying to find published survey results relating to your

target audience: that way, again, you can get a feel for whether your survey

respondents are representative of your target audience, or whether you have

to "interpret" the survey results to extrapolate them. In this way you can

add significant value to your basic online survey activity.


Hope this is of some help.


Regards


Silas Denyer

e.

silas [dot] denyerturns [dot] net

w. www.turns.net



~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

  The biggest, best Chinwag Networking party...ever...

    **27th September 2000**  put it in your diaries now

 RSVP to http://www.chinwag.com/uk-netmarketing/party.shtml

  To sponsor the event email

spareanychangechinwag [dot] com

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

  To unsubscribe or change your list settings go to

http://www.chinwag.com/uk-netmarketing or

helpchinwag [dot] com